Hi Barry, Thank you for pointing this issue. I think I will try Elemental for parallel dense matrix, otherwise I will stick on PETSc ;-)
Cheers Gao ________________________________________ From: petsc-users-bounces at mcs.anl.gov [petsc-users-bounces at mcs.anl.gov] on behalf of Barry Smith [[email protected]] Sent: Thursday, April 05, 2012 5:29 PM To: PETSc users list Subject: Re: [petsc-users] question about MatMatMultTranspose On Apr 5, 2012, at 10:27 AM, Gao Bin wrote: > Hi, again > > Sorry, I was wrong in my last email about the interface to PLAPACK in PETSc. > It looks like I could do parallel dense matrix multiplication using PETSc by > enabling PLAPACK interface. There are problems with PLAPACK, even the matrix-matrix product has buggyness issues. I don't recommend you do this. Barry > > Cheers > > Gao > ________________________________________ > From: petsc-users-bounces at mcs.anl.gov [petsc-users-bounces at mcs.anl.gov] > on behalf of Barry Smith [bsmith at mcs.anl.gov] > Sent: Thursday, April 05, 2012 3:02 PM > To: PETSc users list > Subject: Re: [petsc-users] question about MatMatMultTranspose > > Gao, > > PETSc is mostly designed and implemented for large sparse matrix problems. > We are not really experts for large dense matrix problems. Note that PETSc > Seq dense matrices are just stored using the usual column oriented single > array for the matrix (like Blas 2 and 3) so you can always use MatGetArray() > and make some dense computations yourself directly. > > Parallel dense we know very little about and cannot write those routines, > sadly there are no decently supported parallel dense matrix general purpose > libraries out there that we can use (and no Scalapack, plapack and elemental > do not count as decent AND supported) so it is unlikely WE will write the > MPIDENSE versions of these routines. Though if someone else writes them we > would be happy to include them. So basically for parallel dense I have no > suggestions. > > Barry > > On Apr 5, 2012, at 7:42 AM, Gao Bin wrote: > >> Hi Jed, >> >> Good to know it is simpler ;-) I am switching to the developed version and >> try it. Again, thank you very much. >> >> P.S., Moreover, I notice that some functions is not for MATMPIDENSE. May I >> ask if they are too difficult to implement (for instance, C=A*B^T and >> C=A^T*B for MATMPIDENSE)? Thank you. >> >> Cheers >> >> Gao >> From: petsc-users-bounces at mcs.anl.gov [petsc-users-bounces at >> mcs.anl.gov] on behalf of Jed Brown [jedbrown at mcs.anl.gov] >> Sent: Thursday, April 05, 2012 2:32 PM >> To: PETSc users list; Hong Zhang >> Subject: Re: [petsc-users] question about MatMatMultTranspose >> >> On Thu, Apr 5, 2012 at 05:16, Gao Bin <bin.gao at uit.no> wrote: >> Thank you for your quick reply. But as pointed out at >> http://www.mcs.anl.gov/petsc/petsc-dev/docs/manualpages/Mat/MatMatTransposeMult.html: >> >> This routine is currently only implemented for pairs of SeqAIJ matrices. C >> will be of type MATSEQAIJ. >> >> Therefore I can not use it for dense matrix, am I right? If so, will >> MatMatTransposeMult be extended for other types of matrix later on? Thank >> you very much. >> >> This is much simpler than the sparse case. Hong, did you intend to get >> around to this? >
